Learning Objectives




Warm-up

Describe the qualities of this plot.




Data Visualization using ggplot2

ggplot2 is a plotting package that makes it simple to create complex plots from data in a data frame. It provides a more programmatic interface for specifying what variables to plot, how they are displayed, and general visual properties. Therefore, we only need minimal changes if the underlying data change or if we decide to change from a bar plot to a scatter plot. This helps in creating publication quality plots with minimal amounts of adjustments and tweaking.

Packages in R are basically sets of additional functions that let you do more stuff. The functions we’ve been using so far, like str() or mean(), come built into R; packages give you access to more of them. Before you use a package for the first time you need to install it on your machine, and then you should import it in every subsequent R session when you need it. You should already have installed the tidyverse package. This is an “umbrella-package” that installs several packages useful for data analysis which work together well such as tidyr, dplyr, ggplot2, readr, forcats, etc.

The tidyverse package tries to address common issues that arise when doing data analysis with some of the functions that come with R.

  1. The tidyverse solves complex problems by combining many simple pieces.

No matter how complex and polished the individual operations are, it is often the quality of the glue that most directly determines the power of the system. - Hal Abelson

  1. The tidyverse is written for people to read!

Computer efficiency is a secondary concern because the bottleneck in most data analysis is thinking time, not computing time. - Hadley Wickham

If we haven’t already done so, we can type install.packages("tidyverse") straight into the console. In fact, it’s better to write this in the console than in our script for any package, as there’s no need to re-install packages every time we run the script.

Then, to load the package type:

## load the tidyverse packages
library(tidyverse)

To learn more about ggplot2 after the workshop, you may want to check out this ggplot2 reference website (link) and this handy cheatsheet on ggplot2 (link).

Data for the Workshop

We have seen in our previous lesson that when building or importing a data frame, the columns that contain characters (i.e., text) are coerced (=converted) into the factor data type. We had to set stringsAsFactors to FALSE to avoid this hidden argument to convert our data type.

This time we will use the readr package (from the tidyverse) to read in the data and avoid having to set stringsAsFactors to FALSE

We’ll read in our data using the read_csv() function instead of the read.csv() function we used in the Introduction to R workshop.

surveys <- read_csv("data/surveys.csv")
## Parsed with column specification:
## cols(
##   record_id = col_double(),
##   month = col_double(),
##   day = col_double(),
##   year = col_double(),
##   plot_id = col_double(),
##   species_id = col_character(),
##   sex = col_character(),
##   hindfoot_length = col_double(),
##   weight = col_double(),
##   date = col_date(format = ""),
##   day_of_week = col_character(),
##   plot_type = col_character(),
##   genus = col_character(),
##   species = col_character(),
##   taxa = col_character()
## )

You will see the message Parsed with column specification, followed by each column name and its data type. When you execute read_csv on a data file, it looks through the first 1000 rows of each column and guesses the data type for each column as it reads it into R. For example, in this dataset, read_csv reads weight as col_double (a numeric data type), and species as col_character.

## inspect the data
str(surveys)
## preview the data
View(surveys)

Notice that the class of the data is now tbl_df

This is referred to as a “tibble”. Tibbles tweak some of the behaviors of the data frame objects we introduced in the previous workshop. The data structure is very similar to a data frame, so for our purposes the only differences are that:

  1. In addition to displaying the data type of each column under its name, it only prints the first few rows of data and only as many columns as fit on one screen.
  2. Columns of class character are never converted into factors.

Plotting with ggplot2

ggplot2 functions like data in the ‘long’ format, i.e., a column for every dimension, and a row for every observation. There are other data formats, which we will discuss in the Data Wrangling in R workshop, as well as how to covert from one data format to another. Well-structured data will save you lots of time when making figures with ggplot2 and when working in R!

ggplot() graphics are built step by step by adding new elements. Adding layers in this fashion allows for extensive flexibility and customization of plots.

To build a ggplot(), we will use the following basic template that can be used for different types of plots:

ggplot(data = <DATA>, mapping = aes(<VARIABLE MAPPINGS>)) + <GEOM_FUNCTION>()


Let’s go through this step by step!

  1. Use the ggplot() function and bind the plot to a specific data frame using the data argument
ggplot(data = surveys) 

## Creates a blank ggplot(), referencing the surveys dataset
  1. Define a mapping (using the aesthetic (aes) function), by selecting the variables to be plotted and specifying how to present them in the graph, e.g. as x/y positions or characteristics such as size, shape, color, etc.
ggplot(data = surveys, 
       mapping = aes(x = weight, y = hindfoot_length))

## Creates a blank ggplot(), with the variables mapped to the x- and y-axis
## ggplot() knows where the variables live, since you have defined the data to use

  1. Add ‘geoms’ – graphical representations of the data in the plot (points, lines, bars). ggplot2 offers many different geoms; we will use some common ones today, including:

    • geom_point() for scatter plots, dot plots, etc.
    • geom_boxplot() for boxplots
    • geom_bar() for bar charts
    • geom_line() for trend lines, time series, etc.

To add a geom to the plot use the + operator. Because we have two continuous variables in the data, let’s use geom_point() first:

ggplot(data = surveys, 
       mapping = aes(x = weight, y = hindfoot_length)) +
  geom_point()

 ## Adds a point for each row (observation) in the data

You can think of the + sign as adding layers to the plot. Each + sign must be placed at the end of the line containing the previous layer. If, instead, the + sign is added at the beginning of the line containing the new layer, ggplot2 will not add the new layer and will return an error message.

# This is the correct syntax for adding layers
surveys_plot +
  geom_point()

# This will not add the new layer and will return an error message
surveys_plot
  + geom_point()

Building Plots Iteratively

Building plots with ggplot2 is typically an iterative process. We start by defining the dataset we’ll use, lay out the axes, and choose a geom:

ggplot(data = surveys, 
       mapping = aes(x = weight, y = hindfoot_length)) +
    geom_point()

Then, we start modifying this plot to extract more information from it. For instance, we can add transparency (alpha) to the points, to avoid overplotting:

ggplot(data = surveys, 
       mapping = aes(x = weight, y = hindfoot_length)) +
    geom_point(alpha = 0.1)

    ## alpha reduces the opacity of the points 
    ## 0 is fully transparent
    ## 1 is the original opacity

We can also add colors for all the points:

ggplot(data = surveys, 
       mapping = aes(x = weight, y = hindfoot_length)) +
    geom_point(alpha = 0.1, color = "blue")

Challenge 1

geom_point also accepts aesthetics of size and shape. The size of a point is its width in mm. The shape of a point has five different options for plotting:

  • an integer [0, 25] of defined plotting characters – same as base R
  • the name of the shape in quotations (e.g. “circle open” or “diamond filled”)
  • a single character, to use that character as a plotting symbol
  • a “.” to draw the smallest point that is visible – typically 1 pixel
  • an NA, to draw nothing

Reference for shapes in integers and characters:
https://ggplot2.tidyverse.org/articles/ggplot2-specs.html

Modify the previous code chunk to assign one of these aesthetics to the geom_point aspect of your plot. What happened?

## Your ggplot code to answer the challenge goes here!

Piping Data In

Because ggplot2 lives in the tidyverse, it is expected to work well with other packages in the tidyverse. Because of this, the first argument to creating a ggplot() is the dataset you wish to be working with. The pipe operator sends the output of one function directly into the next function, which is useful when you need to do many things to the same dataset. Since the dataset we wish to use is the first argument to ggplot(), we can use the pipe operator to pipe the data into the ggplot() function!

Pipes in R look like %>% and are made available via the magrittr package, installed automatically with the tidyverse. If you use RStudio, you can type the pipe with Ctrl + Shift + M if you have a PC or Cmd + Shift + M if you have a Mac.

This would instead look like this:

surveys %>%  
  ## data to be used in the ggplot  
  ggplot(mapping = aes(x = weight, y = hindfoot_length)) + 
  ## uses the data piped in as the first argument to ggplot() 
    geom_point(alpha = 0.1, color = "blue")

Once we pipe the data in, the first argument becomes the mapping of the aesthetics. Technically, we are using the name of this argument, which is why it looks like:

mapping = aes(<VARIABLES>)

When we pipe our data in, the first argument then becomes this mapping argument.

Assigning More Variables to Aesthetics

Or to color each species in the plot differently, you could use a vector as an input to the argument color. ggplot2 will provide a different color corresponding to different values in the vector. Here is an example where we color with species_id:

surveys %>% 
  ggplot(mapping = aes(x = weight, y = hindfoot_length)) +
    geom_point(alpha = 0.1, aes(color = species_id))

Note: When specifying an alpha for a scatterplot, it automatically uses that same alpha in the legend. To remedy this you can add:

guides(colour = guide_legend(override.aes = list(alpha = 1)))

to your plot. This customizes the legend appearance, similar to what we will see in the customization section.



We can also specify the colors directly inside the mapping provided in the ggplot() function. This will be seen by any geom layers and the mapping will be determined by the x- and y-axis set up in aes().

surveys %>% 
  ggplot(mapping = aes(x = weight, y = hindfoot_length, color = species_id)) +
    geom_point(alpha = 0.1)

Notice that we can change the geom layer and colors will be still determined by species_id

Local Aesthetics versus Global Aesthetics

When you define aesthetics in the ggplot() function, those mappings hold for every aspect of your plot.

For example, if you chose to add a trend line to your plot of weight versus hindfoot length, you would get different lines depending on where you define your color aesthetics.

Globally

surveys %>% 
  ggplot(mapping = aes(x = weight, y = hindfoot_length, color = species_id)) +
  geom_jitter(alpha = 0.1) + 
  geom_smooth()
## `geom_smooth()` using method = 'gam' and formula 'y ~ s(x, bs = "cs")'

  ## trend line for each species_id -- because color is defined globally

Locally

surveys %>% 
  ggplot(mapping = aes(x = weight, y = hindfoot_length)) +
  geom_jitter(aes(color = species_id), alpha = 0.1) + 
  geom_smooth()
## `geom_smooth()` using method = 'gam' and formula 'y ~ s(x, bs = "cs")'

  ## one trend line -- no color defined globally

Challenge 2

Part 1: Inspect the geom_point help file to see what other aesthetics are available. Map a new variable from the dataset to another aesthetic in your plot. What happened? Does the aesthetic change if you use a continuous variable versus a categorical/discrete variable?

## Your ggplot() code for the challenge goes here!

Part 2: Use what you just learned to create a scatter plot of weight over plot_id with data from different plot types being showed in different colors. Is this a good way to show this type of data?

## Your ggplot() code for the challenge goes here!

Boxplots & Violin Plots

We can use boxplots to visualize the distribution of weight within each species:

surveys %>% 
  ggplot(mapping = aes(x = species_id, y = weight)) +
    geom_boxplot()

By adding points to boxplot, we can have a better idea of the number of measurements and their distribution:

surveys %>% 
  ggplot(mapping = aes(x = species_id, y = weight)) +
    geom_boxplot(alpha = 0) + 
    ## alpha = 0 eliminates the black (outlier) points, so they're not plotted twice
    geom_jitter(alpha = 0.2, color = "tomato")

    ## alpha = 0.2 decreases the opacity of the points, to not be too busy

Notice how the boxplot layer is behind the jitter layer? What would you change in the code to put the boxplot in front of the points?

Challenge 3

Part 1: Boxplots are useful summaries, but hide the shape of the distribution. For example, if the distribution is bimodal, we would not see it in a boxplot. A superior density plot is the violin plot, where the shape (of the density of points) is drawn.

Replace the box plot with a violin plot. For help see geom_violin().

##  Start with the boxplot we created:
ggplot(data = surveys, mapping = aes(x = species_id, y = weight)) +
  geom_boxplot(alpha = 0) +
  geom_jitter(alpha = 0.3, color = "tomato")

##  1. Replace the boxplot with a violin plot. For help, see geom_violin().

Part 2: So far, we’ve looked at the distribution of weight within species. Let’s try making a new plot to explore the distribution of another variable within each species.

Create a boxplot for hindfoot_length. Overlay the boxplot layer on a jitter layer to show actual measurements.

##  First: create boxplot for hindfoot_length` overlaid on a jitter layer.

Now, add color to the data points on your boxplot according to the plot from which the sample was taken (plot_id).

Hint: Check the class for plot_id. If plot_id was a character instead, how would the graph be different?

## Next: add color to the data points on your boxplot according to the
## plot from which the sample was taken (plot_id).


## Hint: Check the class for plot_id`. If plot_id was a character instead, 
## how would the graph be different?

Plotting Single Variables – Quantitative & Categorical

Distribution Plots

If we wish to visualize the distribution of a single quantitative variable, our plot changes a bit. Unfortunately, the geom_violin() function only accepts groups, so we cannot make a violin plot with no groups. Darn it!

But, a violin is simply a density plot that’s been reflected across the y-axis. So, we could likely suffice with a density plot.

To visualize the distribution of rodent weights we could aggregate over all species, years, plots, etc. and produce a single density plot:

surveys %>% 
  ggplot(mapping = aes(x = weight)) + 
  geom_density()

The default is an empty density plot, which is largely unsatisfying. By adding a fill = <COLOR> argument to geom_density() we can produce a nicer looking plot:

surveys %>% 
  ggplot(mapping = aes(x = weight)) + 
  geom_density(fill = "sky blue")

Another frequently used plot for a single quantitative variable is the histogram. The same plot as above can be recreated using geom_histogram() instead of geom_density(). However, when you use geom_histogram() it gives you a warning.


What warning do you get and why? Do you get an error like this when you use hist() in base R?


surveys %>% 
  ggplot(mapping = aes(x = weight)) + 
  geom_histogram(fill = "sky blue")

I was once told that the idea behind a good histogram was to make the plot look as smooth as possible – think trying to resemble the continuous shape of the density plot.


Challenge 4

Use the bins argument to play around with the number of bins in your histogram. Compare your chosen number of bins with your neighbors!

Bar Charts

At first glimpse, you would think that a bar plot would be simple to create, but bar plots reveal a subtle nuance of the plots we have created thus far. The following bar chart displays the total number of rodents in the surveys dataset, grouped by their species ID.

surveys %>% 
  ggplot(mapping = aes(x = species_id)) + 
  geom_bar()

The x-axis displays the levels of species_id, a variable in the surveys dataset. On the y-axis count is displayed, but count is not a variable in our dataset! Where did count come from? Graphs, such as the scatterplots, display the raw values of your data. Other graphs, like bar charts and boxplots, calculate new values (from your data) to plot.

  • Bar charts and histograms bin your data and then plot the number of observations that fall in each bin.

  • Boxplots find summaries of your data (min, max, quantiles, median) and plot those summaries in a tidy box, with “outliers” (data over 1.5*IQR from min/max) plotted as points.

  • Smoothers (as used in geom_smooth) fit a model to your data (you can specify, but we used the gam default) and then plot the predictions from that model (with associated confidence intervals).

To calculate each of these summaries of the data, R uses a different statistical transformation, or stat for short. With a bar chart this looks like the following process:

  1. geom_bar first looks at the entire data frame
  2. geom_bar then transforms the data using the count statistic
  3. the count statistic returns a data frame with the number of observations (rows) associated with each level of species_id
  4. geom_bar uses this summary data frame, to build the plot – levels of species_id are plotted on the x-axis and count is plotted on the y-axis

Generally, you can use geoms and stats interchangeably. This is because every geom has a default stat and visa versa. For example, the following code produces the same output as above:

surveys %>% 
  ggplot(mapping = aes(x = species_id)) + 
  stat_count()

If you so wish you could override the default stat for that geom. For example, if you wanted to plot a bar chart of proportions you would use the following code to override the count stat:

surveys %>% 
  ggplot(mapping = aes(x = species_id)) + 
  geom_bar(aes(y = ..prop.., group = 1))



Challenge 5

Why do we need to set group = 1 in the above proportion bar chart? In other words, what is wrong with the plot below?



surveys %>% 
  ggplot(mapping = aes(x = species_id)) + 
  geom_bar(aes(y = ..prop..))

Colored and/or Stacked Bar Charts

Another piece of visual appeal to creating a bar chart is the ability to use colors to differentiate the different groups, or to plot two different variables in one bar chart (stacked bar chart). Let’s start with adding color to our bar chart.

As we saw before, to add a color aesthetic to the plot we need to map it to a variable. However, if we use the color option that we used before we get a slightly unsatisfying result.

surveys %>% 
  ggplot(mapping = aes(x = species_id, color = species_id)) + 
  geom_bar()

We notice that the color only appears in the outline of the bars. For a bar chart, the aesthetic that we are interested in is the fill of the bars.

Challenge 6

Change the above code so that each bar is filled with a different color.


Now suppose you are interested in whether the number of rodents in each species captured differs by sex. This would require for you to create a bar plot with two categorical variables. You have two options:

  1. each of the bars for sex could be stacked within a species OR
  2. the bars for sex could be side-by-side within a species

Let’s see how the two approaches differ. To stack bars of a second categorical variable we would instead use this second categorical variable as the fill of the bars. Run these two lines of code and see how they differ. 

surveys %>% 
  ggplot(mapping = aes(x = species_id, fill = sex)) + 
  geom_bar()

surveys %>% 
  ggplot(mapping = aes(x = species_id, fill = sex)) + 
  geom_bar(position = "dodge")



In the first plot, the position was chosen automatically, but in the second plot the position argument was made explicit. What changes did this make in the plots?

Plotting Time-series Data

Let’s calculate number of counts per year for each genus.

What you will see in Data Wrangling: First we need to group the data and count records within each group!

yearly_counts <- surveys %>%
  count(year, genus) 
 ## counts the number of observations (rows) for each year, genus combination

Time series data can be visualized as a line plot with years on the x-axis and counts on the y-axis:

yearly_counts %>% 
  ggplot(mapping = aes(x = year, y = n)) +
     geom_line()

Unfortunately, this does not work because we plotted data for all the genera together. We need to tell ggplot() to draw a line for each genus by modifying the aesthetic function to include group = genus:

yearly_counts %>% 
  ggplot(mapping = aes(x = year, y = n, group = genus)) +
    geom_line()

Unfortunately, we can’t tell what line corresponds to which genus. We will be able to distinguish genera in the plot if we add colors (using color also automatically groups the data):

yearly_counts %>% 
ggplot(mapping = aes(x = year, y = n, color = genus)) +
    geom_line()

Note: When specifying the color for a line graph, you can use either the color = <VARIABLE> argument or the group = <VARIABLE> argument. Both do the same grouping of observations!

Faceting

ggplot2 has a special technique called faceting that allows the user to split one plot into multiple plots based on a categorical variable included in the dataset.

There are two types of facet functions:

  • facet_wrap() arranges a one-dimensional sequence of panels to allow them to cleanly fit on one page – used for one variable
  • facet_grid() allows you to form a matrix of rows and columns of panels – used for two variables

Both geometries allow to to specify faceting variables specified within the vars() function. The vars() function looks at the categorical variable you provide it with and outputs the unique levels (values) of that variable.

This looks like: facet_wrap(facets = vars(facet_variable)) or facet_grid(rows = vars(row_variable), cols = vars(col_variable)).


Let’s start by using facet_wrap() to make a time series plot for each species:

yearly_counts %>% 
  ggplot(mapping = aes(x = year, y = n)) +
    geom_line() +
    facet_wrap(facets = vars(genus))

Now we would like to split the line in each plot by the sex of the rodent captured. To do that we need to make counts in the data frame grouped by year, species_id, and sex:

yearly_sex_counts <- surveys %>%
  count(year, species_id, sex)
## counts the number of observations (rows) for each year, species, sex combination

We can now make the faceted plot by splitting further by sex using color (within each panel):

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
    geom_line() +
    facet_wrap(facets =  vars(species_id))

You can also organize the panels only by rows (or only by columns), using the optional nrow and ncol arguments:

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id), ncol = 1)

  # One column, facet by rows
yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
    geom_line() +
    facet_wrap(vars(species_id), nrow = 1)

  # One row, facet by columns



Now let’s use facet_grid() to control how panels are organized by both rows and columns:

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
    geom_line() +
    facet_grid(rows = vars(sex), cols =  vars(species_id))

Note: In earlier versions of ggplot2 you need to use an interface using formulas to specify how plots are faceted (and this is still supported in new versions). The equivalent syntax is:

# facet wrap
facet_wrap(vars(genus))    # new
facet_wrap(~ genus)        # old

# grid on both rows and columns
facet_grid(rows = vars(genus), cols = vars(sex))   # new
facet_grid(genus ~ sex)                            # old

# grid on rows only
facet_grid(rows = vars(genus))   # new
facet_grid(genus ~ .)            # old

# grid on columns only
facet_grid(cols = vars(genus))   # new
facet_grid(. ~ genus)            # old

Challenge 7

Use what you just learned to create a plot that depicts how the average weight of each species changes through the years. Play around with which variable you facet by versus plot by!

## To get you started:  
yearly_species_weight <- surveys %>% 
  group_by(year, species_id) %>%
  ## Variables to group by!
  summarize(avg_weight = mean(weight))
## `summarise()` regrouping output by 'year' (override with `.groups` argument)
## Your ggplot() code for the plot goes here!

ggplot2 Themes

Usually plots with white background look more readable when printed. Every single component of a ggplot() graph can be customized using the generic theme() function, as we will see below. However, there are pre-loaded themes available that change the overall appearance of the graph without much effort.

For example, we can change our previous graph to have a simpler white background using the theme_bw() function:

yearly_sex_counts %>% 
 ggplot(mapping = aes(x = year, y = n, color = sex)) +
     geom_line() +
     facet_wrap(vars(species_id)) +
     theme_bw()

In addition to theme_bw(), which changes the plot background to white, ggplot2comes with several other themes which can be useful to quickly change the look of your visualization. The complete list of themes is available at https://ggplot2.tidyverse.org/reference/ggtheme.html. theme_minimal() and theme_light() are popular, and theme_void() can be useful as a starting point to create a new hand-crafted theme.

The ggthemes package provides a wide variety of options. The ggplot2 extensions website provides a list of packages that extend the capabilities of ggplot2, including additional themes.

Challenge 8

Use what you just learned to add the plotting background theme of your choosing to the plot you made in Challenge 7!

## Your ggplot() code for the plot goes here!

Customization

Take a look at the ggplot2 cheat sheet, and think of ways you could improve the plot.

Plot Labels

Now, let’s change names of axes to something more informative than ‘year’ and ‘n’ and add a title to the figure. Label customizations are done using the labs() function like so:

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id)) +
  theme_bw() +
  labs(title = "Observed Genera Through Time",
       x = "Year of Observation",
       y = "Number of Rodents",  
       color = "Sex")

Tip: Wrapping Titles
Sometimes the titles we wish to have for our plots are longer than the space originally allotted. If you create a title and the text is running off the plot you can add a \n inside your title to force a line break (\n stands for new line).

Label & Plot Fonts

Note that it is also possible to change the fonts of your plots. If you are on Windows, you may have to install the extrafont package, and follow the instructions included in the README for this package.

In the last plot, the axes have more informative names, but their readability can be improved by increasing the font size. This can be done with the generic theme() function.

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id)) +
  theme_bw() +
  labs(title = "Observed Genera Through Time",
      x = "Year of Observation",
      y = "Number of Rodents", 
      color = "Sex") +
  theme(text = element_text(size = 16))

    ## sets ALL the text on the plot to be size 16

Note:
theme_bw() is function for a specific theme and theme() is a generic function for a variety of different themes!


After our manipulations, you may notice that the values on the x-axis are still not properly readable. Let’s swap the orientation of the labels, so the reader doesn’t have to tilt their head when reading our plot! The coord_flip() function easily changes the x- and y-axis.

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id)) +
  theme_bw() +
  labs(title = "Observed Genera by \n Year of Observation",
      x = "",
      y = "Number of Rodents", 
      color = "Sex") +
  theme(text = element_text(size = 16)) + 
  coord_flip()

This definitely makes the reader tilt their head less! But, the text on the x-axis is a bit too large to separate the numbers. We can specify the text size for each element of the plot independently, if we so wish. This would look something like this:

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id)) +
  theme_bw() +
  labs(title = "Observed Genera by Year of Observation",
        x = "",
        y = "Number of Rodents", 
        color = "Sex") +
  theme(axis.text.x = element_text(size = 10), 
        axis.text.y = element_text(size = 12), 
        axis.title.x = element_text(size = 14),
        legend.text = element_text(size = 12),
        legend.title = element_text(size = 12),
        plot.title = element_text(size = 16)) + 
  coord_flip()

Legend Position

By default in ggplot2 the legend is positioned on the right hand side. However, you are able to change the position of the legend to the left hand side, the top of the plot, or the bottom of the plot.

This is done by adding a legend.position theme to the plot’s theme()’s.

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
    geom_line() +
    facet_wrap(vars(species_id)) +
    labs(title = "Observed Genera by Year of Observation by",
        x = "",
        y = "Number of Rodents", 
        color = "Sex") +
  theme_bw() +
  theme(axis.text.x = element_text(size = 10), 
        axis.text.y = element_text(size = 12), 
        axis.title.x = element_text(size = 14),
        legend.text = element_text(size = 12),
        legend.title = element_text(size = 14),
        plot.title = element_text(size = 14), 
        legend.position = "top") + 
  coord_flip()

Removing Grid Lines

By default, the background of a ggplot() contains both minor and major gridlines. These can make the plot look a bit busy and difficult for the reader to follow. As you may have guessed, to remove these gridlines, we add another theme to our plot.

This looks like this:

yearly_sex_counts %>% 
  ggplot(mapping = aes(x = year, y = n, color = sex)) +
  geom_line() +
  facet_wrap(vars(species_id)) +
  labs(title = "Observed Genera by Year of Observation by",
      x = "",
      y = "Number of Rodents", 
      color = "Sex") +
  theme(axis.text.x = element_text(size = 10), 
        axis.text.y = element_text(size = 12), 
        axis.title.x = element_text(size = 14),
        legend.text = element_text(size = 12),
        legend.title = element_text(size = 14),
        plot.title = element_text(size = 14), 
        legend.position = "top", 
        ## New themes for the grid lines
        axis.line = element_line(colour = "black"),
        ## 
        panel.grid.major = element_blank(),
        panel.grid.minor = element_blank(),
        panel.border = element_blank(), 
        panel.background = element_blank()) + 
  coord_flip()

Let’s break these options down!

  • The axis.line option declares what color the x- and y-axis lines should be. (Change it to a different color, if you don’t believe me!)
  • The panel.grid.major removes the major grid (the one associated with the ticks from the x- and y-axis).
  • The panel.grid.minor removes the minor grid (the one between the x- and y-axis ticks).
  • The panel.border removes the border around the plot.
  • The panel.background performs a similar action to theme_bw(), but it keeps the border around the facet labels.

Changing Colors

The built in ggplot() color scheme may not be what you were looking for, but don’t worry! There are many other color palettes available to use!

You can change the colors used by ggplot() a few different ways.

Manual Specification

Add the scale_color_manual() or scale_fill_manual() functions to your plot and directly specify the colors you want to use. You can either:

  • defining a vector of colors right there (e.g. values = c("blue", "black", "red", "green"))

  • creating a vector of colors and storing it in an object and calling it (see below)

# A color deficient friendly palette with grey:
cbPalette_grey <- c("#999999", "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#D55E00", "#CC79A7")

# A color deficient friendly palette with black:
cbPalette_blk <- c("#000000", "#E69F00", "#56B4E9", "#009E73", "#F0E442", "#0072B2", "#D55E00", "#CC79A7")

surveys %>% 
  ggplot(aes(x = species_id, y = hindfoot_length, color = genus)) +
  geom_boxplot() +
  scale_color_manual(values = cbPalette_grey)

Package Specification

Install a package and use it’s available color scales. Popular options include:

  • RColorBrewer: using scale_fill_brewer() or scale_colour_brewer()

  • viridis: using scale_colour_viridis_d() for discrete data, scale_colour_viridis_c() for continuous data, with an inside argument of option = <COLOR> for your chosen color scheme

  • ggsci: using scale_color_<PALNAME>() or scale_fill_<PALNAME>(), where you specify the name of the palette you wish to use (e.g. scale_color_aaas())

Challenge 9

With all of this information in hand, please take another five minutes to either improve one of the plots generated in this exercise or create a beautiful graph of your own. Use the RStudio ggplot2 cheat sheet for inspiration. Here are some ideas:

Arranging and Exporting Plots

Faceting is a great tool for splitting one plot into multiple plots, but sometimes you may want to produce a single figure that contains multiple plots using different variables or even different data frames. The gridExtra package allows us to combine separate ggplots into a single figure using grid.arrange():

library(gridExtra)

spp_weight_boxplot <- surveys %>% 
  ggplot(aes(x = genus, y = weight)) +
  geom_violin() +
  geom_jitter(color = "tomato", width = 0.2, alpha = 0.1) + 
  scale_y_log10() +
  ## log (base 10) transforms the y-axis variable
  ## (helps to make the plot less skewed)
  labs(x = "Genus", 
       y = expression(Log[10](Weight))) +
  coord_flip() + 
  theme(axis.text.y = element_text(size = 12), 
        axis.text.x = element_text(size = 12), 
        text = element_text(size = 16))

spp_count_plot <- yearly_counts %>% 
  ggplot(aes(x = year, y = n, color = genus)) +
  geom_line() + 
  labs(x = "Year", y = "Abundance")

grid.arrange(spp_weight_boxplot, spp_count_plot, ncol = 2, widths = c(4, 6))

## nrow and ncol specify how many rows/columns you want the arranged plots to be in
## widths specify what proportion of the overall plotting area each plot takes up

In addition to the ncol and nrow arguments, used to make simple arrangements, there are tools for constructing more complex layouts.

For more assistance arranging plots with grid.arrage(). I find the following vignette very helpful!

https://cran.r-project.org/web/packages/egg/vignettes/Ecosystem.html

Exporting Plots

After creating your plot, you can save it to a file in your favorite format. The Export tab in the Plot pane in RStudio will save your plots at low resolution, which will not be accepted by many journals and will not scale well for posters.

Instead, use the ggsave() function, which allows you easily change the dimension and resolution of your plot by adjusting the appropriate arguments:

Make sure you have the fig/ folder in your working directory.

my_plot <- ggplot(data = yearly_sex_counts, 
                  mapping = aes(x = year, y = n, color = sex)) +
              geom_line() +
              facet_wrap(vars(species_id)) +
              labs(title = "Observed genera through time",
                    x = "Year of observation",
                    y = "Number of individuals") +
              theme_bw() +
              theme(axis.text.x = element_text(colour = "grey20", size = 12,
                                               angle = 90, hjust = 0.5, 
                                               vjust = 0.5),
                    axis.text.y = element_text(colour = "grey20", size = 12),
                    text = element_text(size = 16))

ggsave("fig/yearly_sex_counts.png", my_plot, width = 15, height = 10)

# This also works for grid.arrange() plots
combo_plot <- grid.arrange(spp_weight_boxplot, spp_count_plot, 
                           ncol = 2, widths = c(4, 6))

ggsave("fig/combo_plot_abun_weight.png", combo_plot, width = 10, dpi = 300)

Note: The parameters width and height also determine the font size in the saved plot.




Tips for Creating Visually Appealing Data Visualizations

One of my favorite references when I am creating plots are the slides from Will Chase’s presentation at the 2020 RStudio Conference (link). Will is a graphic designer and lays out some interesting “best practices” for creating graphics that we can explore. His suggestions are for three components of the plot: layout, typography, and color. Here are some big take-aways that I use when I create a plot!












Happy plotting!




Instructor Biographies

Allison Theobold

Allison Theobold, originally from Grand Junction, CO, is a sixth-year graduate student studying Statistics Education at Montana State University. Allison graduated from Colorado Mesa University in 2014, earned a Master’s degree in Statistics from Montana State University, and will be defending her dissertation in April of this year. Recognized by both the Department of Mathematical Sciences and the College of Letters and Sciences as an outstanding graduate teacher and researcher, Allison’s passion for teaching data science is infectious. Allison has taught Introduction and Intermediate Statistics to undergraduate and graduate students, while also completing two years of statistical consulting for diverse groups at Montana State. Her experiences as a teacher and consultant established an interest in preparing researchers in the sciences with the computational tools necessary to implement statistics.



Elijah Meyer

Elijah was born and raised in Great Falls, Montana and currently holds a Master’s degree in Statistics from Montana State University. His early research involved sports statistics, focusing in on Fitbit data and Disc Golf visualizations. His recent work includes research on Graduate Teaching Assistants (GTAs) and program development to help better support GTAs when teaching statistics. Many of these projects were presented to audiences at the Joint Statistical Meetings, Cascadia Symposium of Statistics in Sports, and United States Conference on Teaching Statistics 2019. He is currently pursuing a Ph.D. in statistics with a focus on statistics education.



Sara Mannheimer

Sara Mannheimer is Assistant Professor and Data Librarian at Montana State University, where she helps shape practices and theories for curation, publication, and preservation of data. Her research is rooted in the examination of the social, ethical, and technical issues that arise as a result of our data-driven world.


Workshop Assistant Biographies

Mark Greenwood

Mark Greenwood has been at Montana State University since 2004 and is a Professor of Statistics in the Department of Mathematical Sciences along with Director of Statistical Consulting and Research Services. His research has spanned a wide variety of areas with applications in environmental and clinical areas being main focus areas and methodological work in clustering and functional data analysis being primary areas of research interest. When not teaching or supervising student research and the work done in SCRS, Mark enjoys hiking, road biking, and fly fishing.


Workshop Supporters

These workshops are made possible by the generous support of the Montana State University Library and the Statistical Consulting and Research Services.



Workshop build on: 2020-11-19 17:33:16